Authoritarian Breakdown

Barbara Geddes Department of Political Science

UCLA Los Angeles, CA 90095-1472

Geddes@ucla.edu

January 2004

1

Although most human beings have lived under some form of authoritarianism throughout

most of recorded history, the academic study of politics has focused on democracy. This focus

may reflect the intellectual interests of the predominantly North American and West European

scholars who built the discipline of political science, or it may be due to the greater transparency

and routinization of democratic politics, which makes it easier to observe and to theorize.

Whatever the reason, this focus has left us with few shoulders of giants on which to stand as we

try to understand political change in less democratic and less institutionalized settings.1

The study of regime transition is one of the areas in which the near absence of theories about

authoritarian politics has impeded scholars’ ability to explain events. Since World War II, more

than 125 authoritarian regimes have ended, as have many democracies. Many fine studies of

these transitions have been written, but few of the general explanations proposed by scholars

have turned out to hold across the full range of cases. One reason for the inability to develop a

general explanation for transitions, I argue below, is that different forms of authoritarianism break

down in characteristically different ways. As a consequence, explanations of transition developed

in response to experiences in one part of the world, in which some particular kind of

authoritarianism predominates, offer little leverage for explaining transitions in other regions

where different forms of authoritarianism are more common. Without theories of authoritarian

politics, we perceive these differences as simply more of the great sprawling complexity of

political life, rather than as forming systematic and explicable patterns.

This study contributes to building theories of authoritarian politics. In it, I develop an

argument about intra-elite competition within authoritarian regimes – the analogue of elections in

democracies -- and then draw out some of the argument’s implications for the study of transitions.

I test a hypothesis implied by this argument using a data set of information about nearly all post-

1 Exceptions include the work of Douglass North (1981, 1989, 1990), Margaret Levi (1988), and others who have focused on the struggle between rulers and subjects over property rights and taxation. Juan Linz has long noted the very great differences among different forms of authoritarianism. More recently, Linz, Stepan, and others influenced by their work have tried to develop arguments linking characteristics of authoritarian rule to transition outcomes (Linz and Stepan 1996; Linz and Chehabi 1998; Snyder 1998).

2

1945 authoritarian regimes. One of the clearest findings to emerge from this analysis is that

military governments contain inherent sources of fragility that make them far less durable than

other kinds of authoritarianism. Consequently, they last less long and are more easily

destabilized than are other types of dictatorship.

Past Research on Transitions

The literature on transitions suffers from an odd bifurcation. Only a couple of robust

generalizations have emerged from it, and they both involve relationships between economic

conditions and regime type. These findings are usually interpreted as implying a relationship

between the interests of ordinary citizens and demand for types of political regime. Yet nearly all

detailed descriptions of particular transitions and most efforts to theorize them focus on the

interests, choices, and strategies of elite political actors. It is the propositions suggested in

analyses of elite behavior that seem never to generalize very far beyond the cases they were

developed to explain.2 Scholars seeking to take another step toward understanding transitions

thus face two tasks: to develop general theories of elite behavior in different situations, and to

articulate a more systematic understanding of the interaction between elite decisions and mass

demands.

Perhaps the best established generalization about the effects of economic conditions on

political systems is that people who live in wealthier countries are more likely to enjoy democratic

government. The positive empirical relationship between democracy and economic development

has been well established for some time (Jackman 1973, Bollen 1979), and it has been confirmed

more recently in a series of very sophisticated statistical studies (Burkhart and Lewis-Beck 1994;

Londregan and Poole 1990 and 1996; Przeworski and Limongi 1997; Przeworski et al. 2000).

Londregan and Poole show that the strongest predictor of transitions to authoritarianism, whether

from prior authoritarian or democratic regimes, is poverty. Przeworski and co-authors show that

once democratization has occurred, it survives in countries above a certain level of economic

development. Among countries below that threshold, the likelihood of a reversion to dictatorship

increases with poverty. Earlier scholars saw the relationship between wealth and democracy as

2 See Geddes (1999) for a review of the transitions literature.

3

implying that increasing levels of economic development would lead citizens to demand

democratic governance. Przeworski et al. (2000) argue convincingly, however, that the

relationship is caused not by transitions to democracy in better off countries but rather by the

absence of democratic breakdown in these countries.

A second finding is also reasonably well-established, namely, that poor short-term economic

performance contributes to the downfall of dictatorships, just as it does to the breakdown of

democracies and the defeat of incumbents in stable democracies.3 This finding, unlike the one

above, has a standard interpretation: all governments require some support and even in those

that permit no routine citizen input into leadership choice, those members of society whose

support is needed to maintain leaders in power use a “retrospective voting” calculus in deciding

whether to withdraw their support. Authoritarian governments may be insulated from the distress

of ordinary citizens, but they must deliver benefits to their own, often restricted, group of

supporters in order to survive in power. In the analysis below, I show that some kinds of

authoritarianism are more dependent on economic performance than others.

In contrast to the quantitative studies of the relationship between development and

democracy, most non-quantitative research on transitions from one regime type to another

focuses not on the effect of economic conditions on the likelihood of democracy, but rather on the

interests and strategies of regime and opposition elites and the constraints facing them. Until the

early nineties, one of the most widely accepted generalizations about transitions was that “there

is no transition whose beginning is not the consequence—direct or indirect—of important

divisions within the authoritarian regime itself” (O’Donnell and Schmitter 1986:19). Detailed case

studies of several early Latin American transitions showed that the impetus for the first steps

toward democratization could indeed be found within military governments. Greek experience

followed the same pattern. Studies of the roots of transition in Spain and Portugal highlighted the

factions within those regimes, though claims that these factions actually caused regime change

were less persuasive. The transitions that followed the Soviet collapse, however, could not in

3 Przeworski et al. (2000) understate the effect of economic performance on the survival of authoritarian regimes because they code authoritarian regimes that are followed by other authoritarian regimes as continuing to exist.

4

most cases be traced to splits within the old regime. Nor can most transitions in Africa. Instead,

“transitions in Africa seem to be occurring more commonly from below….” (Bratton and van de

Walle 1997: 83). In short, transitions have occurred in different ways in different parts of the

world.

If, instead of lumping all dictatorships together, we begin the study of transitions with a few

ideas about normal political processes within different kinds of authoritarianism, we can

understand why transitions differ. Military regimes such as those observed by O’Donnell and

Schmitter do, as I show below, contain sources of internal fragility not shared by other types of

authoritarianism.

Types of Authoritarianism

Different kinds of authoritarianism differ from each other as much as they differ from

democracy. Their leaders emerge from different groups and via different selection processes.

They rely on different segments of society for support. They have different procedures for making

decisions, and different interest groups influence policies. Intra-elite factionalism and competition

take different forms in different kinds of dictatorship, and consequently succession occurs in

different ways. They deal with ordinary citizens and opposition in different ways. Because

analysts have not studied these differences systematically, few have considered how different

kinds of competition over succession and policy might affect the likelihood and mode of transition.

These differences, however, cause different kinds of authoritarianism to break down in

systematically different ways. Here I propose theoretical foundations for explaining some of the

differences among types of authoritarianism.

In an effort to capture some of the most important differences among authoritarian regimes, I

classify them as military, single-party, personalist, or amalgams of the pure types.4 In military

regimes, a group of officers decides who will rule and influences policy. In single-party regimes,

one party dominates access to political office and controls policy, though other parties may legally

exist. In personalist regimes, access to office and the fruits of office depends on the discretion of

an individual leader. This leader may wear a uniform and may have created a party to support

4 See Geddes (2003) for criteria used in classifying regimes.

5

himself, but neither the military nor the party exercises independent decision-making power. The

whims of the ruler prevail (cf. Bratton and van de Walle 1997: 61-96; Linz and Chehabi 1998: 4-

45; Snyder 1998).

Factions, which occur in all kinds of government, tend to destabilize military regimes when

conflict between them become intense. Consequently, transitions from military rule usually begin

with splits within the ruling military elite, as noted by much of the literature on Latin American

transitions from military rule. In contrast, rival factions within single-party and personalist regimes

have stronger incentives to hold onto power despite their differences. Single-party regimes are

quite resilient and tend to be brought down by exogenous events rather than competition over

succession or policy disagreements (cf. Haggard and Kaufman 1995; Huntington 1991).

Personalist leaders do everything they can to repress factions. Personalist regimes tend to

collapse in two situations: when the leader dies or becomes quite ill, as that undermines

repressive capacity centralized in his hands; or when disastrous economic events prevent the

distribution of benefits and profit opportunities to close supporters. Because they usually respond

to opposition with repression rather than negotiation, personalist regimes are more likely to end in

bloody upheaval of one kind or another than either of the other types of authoritarianism (cf.

Huntington 1991).

These differences between military and other types of regime explain why observers of

transitions in Africa and Eastern Europe usually find the beginnings of change outside the ruling

clique rather than inside, as would have been expected by the analysts of the earlier transitions in

Latin America. In the rest of this study, I elaborate these arguments and demonstrate that they

are consistent with evidence.

To show why the breakdown of military regimes tends to start from within the officer corps but

the breakdown of other forms of authoritarianism does not, I focus on rivalries and relationships

within the ruling entity of authoritarian governments: the officer corps, dominant party, clique

surrounding the ruler, or some combination of two or more of these. Most of the time, the

greatest threat to the survival of the leader in office – as opposed to the survival of the regime

itself -- comes from within this ruling group, not from outside opposition. In normal times, in other

6

words, most of what we would call politics, the struggle over office, spoils, and policy decisions,

takes place within this ruling group. Thus understanding this struggle plays the same central role

in understanding authoritarian politics that understanding elections does in democracies.

Politics within the ruling group tells only part of the story of regime change, but it is a part that

has been understudied. Opposition from outside the ruling coalition and exogenous shocks, such

as the Soviet collapse, the international economic crisis of the 1980s, and the International

Monetary Fund-induced economic reforms that followed the crisis, have affected regime survival,

sometimes decisively. By focusing on the political dynamics within different kinds of authoritarian

regime, however, I aim to build the foundations for explaining when and how exogenous shocks

and popular mobilization affect different kinds of regime and thus the likelihood of transition.

Having a theoretical basis for understanding different kinds of authoritarian regimes would make it

possible to move beyond the lists of causes-that-sometimes-matter often found in studies of

transitions and toward systematic statements about when particular causes are likely to matter.

Theoretical Foundations

Standard theories of politics begin with two simplifying assumptions: politicians want to

achieve office and remain there; the best strategy for surviving in office is to give those whose

support is needed what they want. Much of the literature on democratic politics concerns how

different political institutions affect the survival strategies of politicians. The analysis of transitions

requires an analogous investigation of the effects of differences among authoritarian institutions,

whether formal or informal. To begin this task, the plausibility of the standard assumptions about

politics needs to be assessed and those assumptions possibly revised. Most obviously, in the

absence of routine ways for citizens to remove leaders from office, questions of who exactly the

supporters of authoritarian leaders are, how satisfied they have to be, and what factors besides

satisfaction with regime performance affect their level of acquiescence require empirical

investigation. These questions cannot be answered in the abstract or assumed away, as in the

study of democratic politics. The identity and interests of regime supporters varies quite

dramatically from one kind of authoritarianism to another.

7

Less obviously, it should not be assumed, for reasons shown below, that the officers, parties,

and cliques supporting authoritarian leaders always want to hold office or that, having achieved it,

they always want to hang onto it. One of the central arguments of this study is that military

officers, in contrast to cadres in single-party and personalist regimes, sometimes do not. If there

are circumstances in which they can achieve their primary goals better out of power than in, as I

will argue there are, then we cannot build theories of authoritarian politics that begin with the

standard assumptions.

What Military Officers Want

As research on the attitudes and preferences of military officers in many different societies

shows, officers in different countries come from different socioeconomic, ethnic, and educational

backgrounds. They feel sympathetic toward different societal groups, and they pursue different

ideologically motivated goals. No assumptions or generalizations can be made about the societal

interests or domestic policies they support. According to the scholarly consensus, however, most

professional officers place a higher value on the survival and effectiveness of the military itself

than on anything else (Janowitz 1960 and 1977; Finer 1975; Bienen 1978; DeCalo 1976).

This interest implies a concern with the maintenance of hierarchy, discipline, and unity within

the military; autonomy from civilian intervention with postings and promotions; and large budgets

to attract high-quality recruits and buy state-of-the-art weapons. Achieving these corporate goals

also increases the likelihood that officers can achieve their individual desires for high salaries and

predictable career advancement. Officers also value the territorial integrity of their nations and

internal order, but the effective pursuit of these goals requires unity, discipline, and adequate

supplies (Stepan 1971; Nordlinger 1977; Barros 1978). In countries in which joining the military

has become a regular path to personal enrichment (for example, Bolivia, Nigeria, Thailand,

Indonesia, and Congo during certain times), acquisitive motives can be assumed to rank high in

most officers’ preferences. They will be most important for some and second or third most

important for others, if only because the continued existence of lucrative opportunities for officers

may depend on the survival of the military as an effective organization. Where individual

acquisitive motives have swamped concern for corporate survival and effectiveness, however, the

8

professionalism of militaries deteriorates, and the officer corps is less likely to serve as a

successful counterweight to ambitious political leaders.

Where corporate interests prevail, such preferences imply that officers agree to join coup

conspiracies only if they believe that the civilian government prevents the achievement of their

main goals, and that many, in fact, will only join if they believe that the military institution itself is

threatened. These preferences are thus consistent with Stepan’s (1971) and Nordlinger’s (1977)

observations about the importance of threats to the military as an institution in the decisions of

officers to join coup conspiracies. As Nordlinger notes, most officers “would probably have much

preferred to remain in the barracks if their objectives, particularly the defense or enhancement of

the military’s corporate interests, could have been realized from that vantage point” (1977: 142).

The worst threat to the military as an institution is civil war in which one part of the armed

forces fights another. Consequently, the most important concern for many officers in deciding

whether to join a coup conspiracy is their assessment of how many other officers will join. What

Nordlinger, Stepan, and others are describing resembles a classic Battle-of-the-Sexes game in

which one member of a couple prefers going to a movie and the other prefers the symphony, but

both would prefer doing something together to doing something alone. The best outcome for

either one always depends on what the other chooses. For officers, the logic of decisions about

seizing power or returning to the barracks is the same. Some officers hanker after political power

and seize any excuse to intervene. Others have internalized legalist values that preclude

intervention except in the most dire circumstances. Most have views that place them somewhere

in between these extremes. Almost all, however, care most about the survival and effectiveness

of the military, and thus they want the military to move either in or out of power as a cohesive

whole.

Figure 1, Panel A depicts this set of preferences as a game. In the particular game shown,

the majority prefers to remain in the barracks. The pay-offs for remaining in the barracks are

shown in the lower right cell. The upper left cell shows the pay-offs for a successful intervention

carried out by a united military. The minority is better off than it was in the barracks, but the

majority is slightly worse off.

9

The minority would prefer to intervene, but would be far worse off if they initiated an

unsuccessful intervention without support from the majority than if they remained unhappily in the

barracks. (Pay-offs for this outcome are shown in the upper right cell.) Participants in an

unsuccessful coup attempt face possible demotion, discharge, court martial, and execution for

treason. The majority faction that opposed the coup is also damaged by the attempt, since the

armed forces will have been weakened, and the government is likely to respond with greater

oversight, reorganization, and interference with promotions and postings to try to insure greater

future loyalty, all of which reduce military autonomy.

The final possible outcome is a successful coup carried out despite minority opposition.

(Pay-offs are shown in the lower left cell.) In this event, the minority that remains loyal to the

ousted civilian government is likely to face the same costs as unsuccessful conspirators:

demotion, discharge, exile, prison, death. The plotters may achieve power, but a weakened

military institution reduces their chances of keeping it. Future conspiracies supported by those

demoted or discharged after the coup become more likely. Once factions of the military take up

arms against each other, it takes years or decades to restore unity and trust.

This is a coordination game: once the military is either in power (upper left cell) or out of

power (lower right cell), neither faction can improve its position unilaterally. Each faction must

have the other’s cooperation in order to secure its preferred option. When the military is out of

power, even if the majority comes to believe it should intervene, it cannot move to its preferred

alternative without cooperation from the minority.

Where interventionists have broad public support, as in Brazil before the 1964 coup or

Argentina before the 1976 coup, and democracy makes plotting relatively safe, extensive

consultation among officers and between officers and civilian allies often precedes coups.

Plotters may delay the seizure of power until the officer corps has reached consensus about the

need for intervention and developed elaborate rules for consultation and leadership rotation.

These consultations aim to insure the cooperation of all major factions in the intervention. Such

10

elaborate efforts to achieve consensus and hammer out post-intervention rules have been

described by observers of a number of interventions.5

Where those who favor intervention lack popular support or a more repressive political

system raises the costs of plotting, plotters may choose a different strategy. Conspirators can

keep the plot secret from all but a few key officers and hope that the rest will go along once key

central institutions have been seized. The seizure of these institutions signals to the rest of the

officer corps that an intervention has occurred. Often the presidential palace, garrisons in and

around the capital city, radio and TV stations, the central telephone exchange, and the main

airport will suffice to make this signal credible. Nordlinger (1977) identifies this strategy as most

common in practice. It is a characteristic of games like Battle-of-the-Sexes that the actor who

succeeds in credibly moving first can always get what he or she wants. The partner, for example,

who announces, “I’ve bought tickets for the symphony Friday night. Do you want to come, or

shall I ask someone else?” has made a credible first move. The couple will go to the symphony.

The first-mover strategy fails when the first move is not credible; in the context of coup decisions,

when most officers do not believe that most other officers will go along with the plotters.

The attempted Spanish coup in 1981 exemplifies a failed first mover strategy. Plotters

believed that much of the officer corps would support an intervention, mostly because of the

threat to national integrity posed by the democratic government’s willingness to negotiate with

Basque and Catalan nationalists. The small group of active conspirators believed that once they

had seized control of the Cortes and key installations in Madrid, the rest of the officer corps and

King Juan Carlos would acquiesce in the coup. The evidence available suggests that most of the

officer corps would have gone along if the king had not immediately begun telephoning the

captains-general and other high ranking officers to inform them that he would resist the coup

(Colomer 1995). For some officers, loyalty to the king was stronger than other values and led

them to ally with the king. For others, the king’s unequivocal opposition indicated which position

most other officers were likely to take, and this information led them to resist intervention in order

to end up on the same side. The coup might well have succeeded if the king’s access to

5 For the 1964 Brazilian coup, see Stepan (1971); for the 1976 Argentine coup, see Fontana

11

telephones and television had been blocked and he had thus been unable to signal his

opposition. Josep Colomer quotes one of the erstwhile conspirators as saying: “The next time,

cut the King’s phone line!” (1995: 121).

When the military controls the government, their decision-making logic remains the same.

Most officers will go along with a credible move by one faction back to the barracks. Strong

disagreements among leading officers over how to respond to economic difficulties or which

among them will next occupy the presidency lead to intense factionalization. When this happens,

some officers are likely to prefer returning to the barracks as a way of avoiding institution-

damaging conflict. Observers see splits in the officer corps at the time of the first moves toward

democracy because the concern over divisions within the military causes some factions to prefer

a return to the barracks. Both hardline and softline factions can use the first mover strategy,

however. Military presidents are in an especially good position to make quite credible first moves

heading back toward the barracks, and most officers will go along if they do. Hardliners can also

chance first mover strategies, ousting more moderate military presidents. This carries more risk,

but if the move is credible, again, most of the officer corps will go along.

For the officer who claws his way to the top of the post-coup junta, the game may change

after a successful seizure of power, as it did for General Pinochet. Most other officers always see

their situation as resembling a Battle-of-the-Sexes game, however, even in the most politicized

and factionalized militaries. Repeated coups by different factions, as in Syria prior to 1970 or

Benin (then called Dahomey) before 1972, would not be possible if most of the army did not go

along with the first mover, either in seizing power or in handing it back to civilians.

Cadres in Single-Party Regimes

The preferences of cadres in dominant parties are much simpler than those of officers. Like

democratic politicians, they want to hold office. Some value office because they want to control

policy, some for the pure enjoyment of influence and power, and some for the illicit material gains

that come with office in some countries. They value the survival and unity of their parties as

contributors to their own survival in office, not as a means to other, more important ends.

(1987); for the 1973 Chilean coup, see Valenzuela (1978).

12

In the game between factions in a single party (shown in Panel B of Figure 1), the best

outcome for everyone is for both factions to hold office (pay-offs shown in the upper left cell). The

worst outcome occurs when both are out of power (shown in lower right cell). The upper right cell

shows the pay-offs when the party has lost control of government, but the minority faction still fills

some seats in the legislature as an opposition. The minority pay-off in opposition is lower than

when the party holds power because an opposition has fewer opportunities to exercise influence

or line pockets. In the lower left cell the minority faction is excluded from office, but the party still

rules. If the minority faction is excluded from office but the party continues in power, the minority

continues to receive some benefits, since its policy preferences are pursued and party

connections are likely to bring various opportunities, but members of the excluded faction receive

none of the specific perquisites of office.

Factions form in single-party regimes around policy differences and competition for

leadership, as they do in other kinds of regimes, but everyone is better off if all factions remain in

office. This is why single parties usually try to coopt rival factions rather than exclude them.

Neither faction would be better off ruling alone, and neither would voluntarily withdraw from office

unless exogenous events changed the costs and benefits of cooperating with each other (and

hence changed the game itself).

Members of Personalist Cliques

During and after seizures of power, cliques based on personal loyalty form among the

friends, relatives, and allies who surround all political leaders. As in single-party regimes, factions

form around potential rivals to the leader. In personalist regimes, one individual dominates the

military, state apparatus, and ruling party (if one exists). Because so much power is concentrated

in the hands of one individual, he generally controls the coalition-building agenda. Consequently,

the game between factions in personalist regimes must be depicted as a game tree instead of a

two-by-two matrix in order to capture the leader’s control over the first move. As shown in Figure

2, the leader’s faction has the initiative, choosing to share the spoils and perks with the rival

faction or not. The choice I have labeled “hoard” can be interpreted either as limiting the

13

opportunities and rents available to the rival faction or as excluding some of its members

altogether.

After the leader’s faction has chosen its strategy, the rival faction must decide whether to

continue supporting the regime or not. During normal times, they have strong reasons to

continue. As Bratton and van de Walle note, “Because they face the prospect of losing all visible

means of support in a political transition, they have little option but to cling to the regime, to sink

or swim with it” (1997: 86).

In contrast to single-party regimes, the leader’s faction in a personalist regime may actually

increase benefits to itself by excluding the rival faction from participation. Where the main

benefits of participation in the government come from access to rents and illicit profit

opportunities, benefits to individual members of the ruling group will be higher if they need not be

shared too widely. Thus the leader’s faction should share only enough to keep the minimum

necessary coalition winning. Keeping the predatory group relatively small may also help keep

damage to the economy below the meltdown threshold, and thus increase the likelihood of

regime survival. Hoarding by the leader’s faction is thus likely. If the hoarding is not too

extensive, the rival faction is usually better off continuing to cooperate, and most of the time that

is what they do.

If the rival faction were to withdraw its support and begin to plot the leader’s overthrow, they

would risk life, liberty, and property. The rewards of a successful overthrow would be high, but so

would the costs of detection, betrayal, or defeat. In Figure 2, the uncertainty over the outcome of

plots is shown as a play by Nature. The plot succeeds with probability p, usually a low number,

and fails with probability 1-p. The rival faction decides whether to continue its support for the

leader’s faction by comparing its pay-off for support with its expected pay-off from a plot. Two

considerations thus affect the choice: the benefits being derived from the status quo and the

potential plotter’s assessment of the risk of plotting. As long as the personalist ruler seems

powerful enough to detect plots and defeat coup attempts, the rival faction will probably continue

to cooperate. The leader’s faction has an incentive to reduce the benefits to the rival faction to a

level just above that needed to prevent plotting. This system is stable as long as the ruler can

14

distribute the minimum level of benefits needed to deter plotting and as long as the ruler himself

maintains control over the security and armed forces. The situations in which these conditions

become less likely are discussed below.

The Effect of Cadre Interests on the Likelihood of Breakdown

Factions exist in all governments; the interests described above determine how likely they are

to undermine regime stability. Most military officers go along with continued intervention

regardless of whether military rule becomes institutionalized, the leader concentrates power in his

own hands, or a rival ousts the original leader, because they view their interests as similar to a

Battle-of-the-Sexes game. They will return to the barracks, however, if the officer corps shows

signs of disintegrating into openly competing factions.

The attitudes of ordinary officers are thus a source of fragility in military regimes. When elite

rivalries or policy differences become intense and factional splits become threatening, stepping

down becomes an attractive option for most. For officers, there is life after democracy, as all but

the highest regime officials can usually return to the barracks with their status and careers

untarnished. Prosecutions for human rights violations have been rare in the immediate aftermath

of transitions from military rule.

Policy differences, leadership struggles, and succession crises also occur in single-party

regimes, but except in extraordinary situations, most cadres just keep their heads down and wait

to see who wins. Consequently, even intense factionalization in single-party regimes tends not to

undermine stability. Only when cadres in single-party regimes see their own political survival as

better served by leaving the party than by remaining loyal do they desert – as, for example, when

they can foresee that the ruling party’s days are numbered. Once a transition is underway, these

desertions may hasten it, but they would not occur in ordinary times.

Rival factions in personalist regimes have more to gain than those in single-party regimes

from ousting the ruler, but in normal circumstances, the risks of plotting deter them. Only when

exogenous events such as serious illness reduce the ruler’s ability to detect plots would we

expect to see defections within the ruling clique.

15

These differences explain why early analyses of regime change, which drew their insights

from the transitions from military rule in Latin America, traced the initiation of political reform to

splits within the ruling elite. In other regions, where rule by the professional military is less

common, factions and splits within the elite can always be found but usually do not lead to regime

collapse. Instead, events external to the regime, such as economic crisis, foreign pressure, and

popular protest, play a larger role in bringing about the fall of dictatorships (Haggard and

Kaufman 1995; Huntington 1991; Bratton and van de Walle 1997; Casper and Taylor 1996).

Implications of the Argument

The cadre-interests argument sketched above implies that military regimes are inherently

more fragile than other kinds of authoritarianism since they tend to break down in response to

internal splits, no matter what the cause of the splits. If that is true, we should expect military

regimes, on average, to last less long than other forms of authoritarianism. We might also expect

them to respond more strongly to economic crisis.

In contrast to military officers, the leaders of personalist regimes generally hang onto power

with tooth and claw. In Bratton and van de Walle’s words, “They resist political openings for as

long as possible and seek to manage the process of transition only after it has been forced on

them” (1997: 83). If they are forced, by foreign pressure for example, to negotiate with

opponents, they renege at the first opportunity on the agreements made.6 Military governments

rarely renege on the agreements they make, not because they could not, but because

agreements are made when most officers want to return to the barracks.

According to the cadre-interests argument, most of the military prefers to return to the

barracks in some circumstances, and, even for most of those officers who would prefer to remain

in government, the cost of resuming a more ordinary military career is low. The cost of loss of

office is higher for cadres in a dominant party, but not, on average, devastating. Many prominent

party leaders in the democratic regimes that follow single-party rule are former cadres of the

dominant party. Although the cadres of a single-party regime cannot be expected to desert when

6 Note, for example, the way Mobutu of Zaire (now Congo), Eyadema of Togo, and various other long-ruling African leaders manipulated electoral rules and intimidated opponents after agreeing, under pressure from international donors, to initiate multi-party elections.

16

times are good, if it looks as though the party’s hegemony will soon end, those who think they

possess the skills to make a success of democratic politics can be expected to form or join

opposition parties. Even those who remain in the ruling party to the bitter end need not despair of

life after democratization, however. Many previously dominant parties continue to function as

effective political actors after democratization (cf. Van de Walle and Butler 1999). In fact, they

have achieved executive office in the second free and fair election after democratization in a

number of ex-communist and African countries.

The members of personalist cliques, however, have fewer options. Joining the opposition

prior to a transition can have very high costs, and many who desert the regime must go into exile

in order to remain alive and out of jail. From exile, they may plot and organize, but few who

remain at home are willing to risk public opposition. Those who stick with the regime to the end

are much less likely to find respected places in the post-transition political world than are the

close supporters of single-party and military regimes. The death or physical incapacity of the

leader changes everything, however, since it tends to undermine the effectiveness of the security

apparatus and thus dramatically reduces the risks of plotting.

Like members of personalist cliques, cadres of single-party regimes have few reasons to

desert in normal circumstances. Furthermore, because power is less concentrated in single-party

regimes, they are less vulnerable to the death or illness of leaders. Thus we should expect

single-party regimes to last longer than either military or personalist regimes. Because the

dominant strategy of the ruling coalition in single-party regimes is to coopt potential opposition,

single-party regimes tend to respond to crisis by granting modest increases in political

participation, increasing opposition representation in the legislature, and granting some

opposition demands. They attempt to give the opposition enough to deter them from risky plots

and uprisings while continuing to hang onto power.

In the most common kind of regime crisis, one caused by economic trouble, the ruling elite in

any kind of authoritarian regime divides into intransigents and moderates as they decide how to

respond. In military regimes, that division itself tends to persuade many officers that the time has

come for a return to the barracks. In personalist regimes, the ruling coalition narrows as the

17

intransigents circle the wagons and exclude moderates from access to increasingly scarce spoils.

Former regime moderates may then join the opposition (cf. Bratton and van de Walle 1997).

Ruling parties, however, often attempt to distract citizens from their economic grievances by

granting them modest political rights. This strategy only works sometimes, but it works often

enough to extend the average lifetime of single-party regimes.

Data for Testing the Argument

To test the argument above, I have collected data on all authoritarian regimes (except

monarchies) of three or more years duration that existed between 1946 and 2000, in countries

existing prior to 1990 with a population of more than one million. Authoritarian regimes already in

existence in 1946, such as those in the Soviet Union, Mexico, and Turkey, are included and their

length of time in office is calculated from the time they actually took power. Countries that

became independent after 1945 enter the data set at the time of independence (if authoritarian).

Countries that have achieved independence since 1990 because of the break-up of communist

states have not been included because the inclusion of a fairly large number of countries with

severely truncated regimes might have biased conclusions.

Regimes are defined as sets of formal and informal rules and procedures for selecting

national leaders and policies. Using this definition, periods of instability and temporary

“moderating” military interventions (Stepan 1971) are considered interregna, not regimes. That

is, they are periods of holding customary rules in temporary abeyance, struggle over rules, or

transition from one set of rules to the next. The three-year threshold is simply a way of excluding

such periods from the data set. This cut-off was chosen after considerable empirical investigation

of very short-lived authoritarian interludes because it introduced the least misclassification into

the data. The military governed during most of these interregna. If these periods were included

in the data set, they would increase the strength of the findings I report below.

Regimes were classified as military, single-party, personalist, or hybrids of these categories.

Military regimes were defined as those governed by an officer or retired officer, with the support

of the military establishment and some routine mechanism for high level officers to influence

policy choice and appointments. Single-party regimes were defined as regimes in which the party

18

has some influence over policy, controls most access to political power and government jobs, and

has functioning local-level organizations. Regimes were considered personalist if the leader, who

usually came to power as an officer in a military coup or as the leader of a single-party

government, had consolidated control over policy, recruitment, and the security apparatus in his

own hands, in the process marginalizing other officers' influence and/or reducing the influence

and functions of the party. In the real world, many regimes have characteristics of more than one

regime type. When regimes had important characteristics of more than one pure regime type,

especially when the area specialist literature contained disagreements about the importance of

military and party institutions, I put them in hybrid categories. 7

Findings

The expectation that military regimes, because they have more endogenous sources of

fragility than other types of authoritarianism, survive less long on average is borne out by

evidence. If we consider the life spans of those regimes that had ended by 2000, military regimes

have lasted on average 8.5 years.8 Personalist regimes survived 15 years on average, and

single-party regimes (excluding those maintained by foreign occupation or threat of intervention)

nearly 24 years.9

Although these differences in the average length of different types of regime are quite large,

we cannot be sure that they really reflect differences caused by regime type. Military regimes are

more common in Latin America, where levels of economic development are relatively high, and

personalist regimes are most common in Africa, where economies tend to be less developed. It

might be that the stronger demand for democracy by citizens of more developed countries

accounts for the shorter duration of military regimes. To test for this possibility, I have carried out

7 The data collection and coding are described in Geddes 2003. 8 Reminder: authoritarian interludes lasting less than three years have been excluded from the data set. The military ruled during most of these interludes. If they were included, the average length of military rule would be reduced. Nordlinger, who did not exclude them from his calculations, found that military regimes last five years on average (1977: 139). 9 Regimes maintained in power by direct foreign occupation or the threat of military intervention have been excluded from the calculation of average lifespan here because their longevity depends on events outside domestic control. The excluded regimes are those in Afganistan 1979-93, Bulgaria 1947-90, Cambodia 1979-90, Czechoslovakia 1948-90, German Democratic Republic 1945-90, Hungary 1949-90, and Poland 1947-89. The average length of these regimes is 34 years.

19

statistical tests of the effect of regime type on the probability of regime breakdown, controlling for

level of development, growth rate, region, and other factors expected to affect regime survival.10

Region is used as a quasi-fixed effects estimator. Fixed effects estimators are used to hold

constant aspects of history and culture that might affect the outcome of interest but that cannot be

directly measured.11 A dummy variable for Foreign Imposed is used to control for the effect of

direct military occupation or the credible threat of foreign intervention that keeps some

authoritarian regimes in power.

I use two different models in the analysis to make sure that the results do not depend on

model specification. The first, reported in the first two columns of Table 1, is a hazard model.

Hazard models are used in medical research and other areas to predict the survival of individuals

with certain conditions, given various treatments. Thus this type of model seems appropriate also

10 The measure of level of development used is the natural log of GDP per capita. The measure of growth is change in GDP per capita averaged over the two years preceding the observation year. Averaging over the observation year and the one before that produces the same substantive results. Economic data are from the Penn World Tables 2000. Values for cases included in Penn World Tables 1992 but excluded from the 2000 update (mostly communist countries that no longer exist) are imputed using Amelia. For most countries, the PWT cover 1950-2000, which means that regime years prior to 1951 are excluded from the statistical analysis. In addition, no economic data from Afganistan, Albania, Cambodia, Libya, North Korea, Vietnam, South Vietnam, and South Yemen are available. Scattered country-years are also missing for a number of other countries; a list of missing country-years is available from the author. Since the period covered is quite long and I cannot think of any reason to believe that transitions during the years covered would be different from those in the years before and after, I do not think the years excluded introduce bias into the results. The countries left out of the data set are different, however, from those included. All but Libya had or have single-party or single-party/personalist regimes. Their lifespans are unusually long (average 34.7 years) and nearly all of them are very poor. If they were included in the data analysis, they would probably further strengthen the coefficient for the effect of single-party regime but reduce the effect of the level of development on regime stability. 11 Usually country dummy variables are used as fixed effects estimators, but they could not be used to analyze these data because they cause countries with only one regime to be dropped from the analysis. In this data set, half the countries have had only one authoritarian regime, either because one stable regime remained in power for several decades or because the country is usually democratic and had only one postwar authoritarian interlude. More serious than the loss of cases per se, regimes in the single-regime cases are, on average, unusually long lived, and the single-regime cases are especially likely to be single-party regimes. The use of country fixed effects estimators would eliminate 60 percent of the single-party regimes from the analysis.

20

for explaining the survival of a different kind of entity.12 The second two columns of Table 1 show

the results of a Logit Model used to estimate the effects of the same variables.13

Hazard ratios have a simple intuitive interpretation. Ratios above one mean that the

variables associated with them increase the probability of regime collapse. Those between zero

and one mean that the variables reduce the chance of breakdown. Because regime types are

nominal categories, they are entered into the models as dummy variables; for example, if the

regime is military, it is coded “one” and otherwise it is coded “zero.” The left out regime type is

personalist, the middle category in terms of longevity. Thus, the hazard ratios reported should be

interpreted as referring to differences between the probability of breakdown in the type of regime

associated with a particular ratio as compared with that of personalist regimes.

In the first column of Table 1, the hazard ratio for military regime is 3.62, which means that,

all else equal, military regimes are more than three times as likely to break down as personalist.

In column one, the hazard ratio for single-party regimes, 0.33, means that, all else equal, single-

party regimes are only one-third as likely to collapse as personalist.

As can be seen in Table 1, military regimes break down more readily than other types. The

hazard ratios for the military regime variable are above three and statistically significant in both

hazard models, and the logit coefficients are positive and statistically significant. The two

intermediate regime types, military/personalist and hybrid (in which personalist/single-party

regimes predominate since there are very few military/single-party regimes) are not very different

from personalist regimes, so it is unsurprising that their hazard ratios and coefficients have

intermediate values and often fail to achieve statistical significance. Single-party regimes,

however, are about as much more resilient than personalist regimes as military are less. Hazard

ratios for the single-party variable show that they are much less likely to fail than personalist, all

else equal, and logit coefficients are negative. All are statistically significant. Finally, the triple

12 Weibull regression is the appropriate form of hazard model to use when the effect of time is not constant, as is true here. 13 Splines have to be added to the logit model to capture the effect of the time trend. They are jointly significant.

21

hybrid regimes that combine aspects of single-party, personalist, and military regime

characteristics are the strongest of all in all models.14

The control variables used in the models also show some interesting effects. As the level of

development rises, authoritarian regimes, like democratic, become more stable. This finding is

consistent with that of Londregan and Poole (1990 and 1996), who find that the best predictor of

coups, in both democratic and authoritarian regimes, is poverty. Like Przeworski et al.’s (2000)

findings, it challenges conventional demand-centered explanations of the relationship between

increased development and democracy. This finding is inconsistent with the idea that the citizens

of more affluent countries are more likely to demand democratization. Rather, it suggests that

when authoritarian governments manage the economy well over the long term, regime allies

remain loyal and citizens supportive or at least acquiescent. That interpretation is reinforced by

the effect of growth on the probability of regime breakdown. In other words, both long- and short-

term economic performance aid authoritarian stability.

In light of the arguments about the effects of religion, culture and colonial heritage on the

development of democratic values, it is somewhat surprising that most of the region variables

show no effect. The left out region here is southern Europe (Portugal, Spain, and Greece), and

we might have expected the regions most culturally distinct from Europe to exhibit differences in

the likelihood of regime transition. Interestingly, the only region with large, statistically significant

hazard ratios and logit coefficients close to significance is South America, the region most

culturally similar to Southern Europe.

Authoritarian regimes in South America seem more likely to break down, even after

controlling for regime type and economic performance. My interpretation for this finding, though it

cannot be proven using these data, is that democratic values are quite widespread in South

America. These values may sometimes be overwhelmed by fear of chaos or economic disaster,

and at these times many support military intervention. But opinion changes when things calm

down or the authoritarian government shows that it is no better at holding economic disaster at

14 The triple hybrid category includes six long-lived and hard to classify regimes. It is included more to prevent these regimes from artificially inflating the effect of other regime types than to draw conclusions about these regimes per se.

22

bay than the democratic government preceding it. The famously fickle bourgeois allies of the

South American authoritarian governments of the sixties and seventies, for example, quickly

deserted them when new economic problems arose.

It might also seem surprising, given the arguments suggesting an affinity between Islam

and authoritarianism, that dictatorships in the Middle East are no more likely to survive than those

in other areas. This finding might well be different if monarchies had been included in the

analysis, but it suggests that non-monarchical forms of authoritarianism may not be especially

durable in the Middle East. The finding does not imply, however, that democratization is likely.

Rather, in the Middle East one dictatorship usually follows another.

Columns two and four of Table 1 show the effects of controlling for two other factors that

might be expected to affect the longevity of authoritarian regimes, the percent Muslim population

and civil war. Many studies have found that Islamic countries are more likely to have

authoritarian governments, so we might suspect that such governments would be more hardy in

Islamic countries. We might also suspect that civil war would increase the likelihood of

breakdown for any kind of regime. The most important result of introducing these variables is that

the effect of regime type is not affected by their inclusion, which adds to our confidence that types

of authoritarianism really do affect regime durability. Percent Islam is statistically significant in

both models, but suggests that regimes in countries with larger Islamic populations are more

likely to fall, not less as we might have expected.15 Civil war has the expected effect of making

breakdown more likely, though it is not quite statistically significant in either model.16

Since military regimes are more fragile, we might expect them to be more easily destabilized

by economic crisis. To test for that possibility, I redo the logit analysis, this time including

interaction terms for Military Regime Type with Growth and Single-Party Regime Type with

Growth.17 The coefficients for these interaction terms show the effect of changes in the growth

15 This finding also might be reversed by including monarchies in the analysis. 16 Dependence on oil exports was also included in some models, but had no effect itself and did not affect substantive results. Findings not shown here because oil data were unavailable for most country-years. 17 Logit results are shown here because their graphs are easy to interpret, but hazard models produced the same substantive results. An interaction between the Triple Hybrid regime type and

23

rate on the probability of regime collapse in the selected regime type relative to that of the left-out

regimes (Personalist and both types of Hybrids). The coefficient for Growth itself shows the effect

of economic performance on the likelihood of survival for the left-out regime types as a group.

Table 2 shows that Military and Single-Party regimes are both more prone to crisis when the

economy fails than are more personalist dictatorships. Neither coefficient quite reaches

conventional levels of statistical significance, but both show the expected negative effect of

economic decline on regime stability. The coefficient for Military regimes is only a little larger (in

absolute terms) than that for Single-Party, but as Figure 3, Panel C shows, this difference in the

coefficient becomes quite substantial when economic growth drops below zero. In personalist

and hybrid kinds of dictatorship, in contrast, the effect of the economy on the probability of

breakdown is much lower (-0.03) and not statistically significant. Although I cannot yet show that

my interpretation of this is correct, I believe that this finding reflects, on the one hand, the

importance of factors not included here, such as leaders’ deaths and foreign intervention, in

explaining the collapse of personalist regimes. On the other hand, it also reflects the ability of

personalist dictators to continue delivering benefits to their close supporters, and thus maintain

regime support, even when the aggregate economy is declining.

Note that even with regime type and growth interaction terms added to the model, the dummy

variable for military regime continues to show that military regimes are more likely to collapse. In

other words, they have sources of fragility other than their susceptibility to crisis when the

economy is doing poorly. With the exception of Growth Rate, which in this model refers only to

growth in the personalist and hybrid regimes, the other coefficients in the model are largely

unchanged by adding the interaction terms. Foreign Imposed continues to have the opposite sign

from that expected, though it still fails to quite reach statistical significance. Civil War still seems

to increase the likelihood of breakdown, as expected, but is also not quite significant.

Logit coefficients have no simple intuitive interpretation, so their effects are often easier to

understand from graphs. Figure 3 shows the effect of economic performance on the likelihood of

regime breakdown in military and single-party regimes, with other variables in the model set at

growth is also shown in order to prevent these unusual regimes from being lumped into the

24

their means or modes. Panel A shows the effect of economic performance on the probability of

breakdown in military regimes. The center line is the predicted probability of collapse, and the top

and bottom lines show the 90 percent confidence interval around the prediction. (Nearly all cases

fall between plus and minus ten percent change in per capita GDP per year.) The graph shows

that military regimes are fairly unstable when growing at anything below about five percent per

year and that the likelihood of breakdown rises rapidly and linearly as economic performance

declines. Panel B shows the same predicted breakdown rate and 90 percent confidence interval

for single-party regimes. They do not begin to become more unstable until the economy is falling

at about three percent a year, so they have a considerable cushion to get them through hard

times, but when real crisis sets in, the probability of collapse rises. Panel C compares the

predicted probability of breakdown for military and single-party regimes, at any given rate of

growth. As can be seen in this graph, military regimes are more likely than others to fall at any

level of economic performance, but as the economy worsens, their disadvantage relative to

single-party regimes worsens.

Personalist regimes (not shown) are in all economic circumstances less likely to fall than

military regimes and in most, more likely to fall than single-party. Economic performance has a

less predictable effect on them, but they have other sources of fragility relative to single-party

regimes.

Conclusion

This study makes two contributions to the literature on regime transitions: it proposes a

theoretical innovation that subsumes a number of the apparently contradictory arguments made

in the literature; and it tests two implications drawn from this theoretical argument on a data set of

nearly all post World War II authoritarian regimes. The theoretical argument begins with a simple

game-theoretic portrayal of the incentives facing officers in military regimes as contrasted with

those of cadres in single-party regimes and clique members in personalist regimes. If the

incentives shown in the games are, on average, accurate, then we can understand why the

Personalist category.

25

process of transition from military regimes differs from that in single-party and personalist

regimes. Because most officers value the unity and capacity of the military institution more than

they value holding office, they cling less tightly to power than do office holders in other forms of

authoritarianism, and, indeed, often initiate transitions.

This basic insight leads to explanations for many of the differences between early transitions,

mostly from military rule, and later transitions, from personalist and single-party rule. Most

transitions from military rule begin, as O’Donnell and Schmitter (1986) note, with internal

disagreements and splits. Most personalist and single-party regimes, however, maintain their

grip on power as long as possible. Consequently, as the data analysis shows, they last longer.

Because of these differences between officers’ and cadres’ motivations, economic crisis --

usually seen as the most important single reason for regime breakdown – affects different types

of dictatorship differently. Good economic performance strengthens all kinds of regimes, though

it is insufficient to overcome the inherent weakness of military regimes. Poor performance,

however, is devastating for military regimes. They can weather very little economic bad news

and, on average, collapse with higher growth rates than do other types.

Transitions from single-party rule, though the subject of numerous case studies, have played

a lesser role in the comparative transitions literature because fewer have occurred (besides those

caused by the Soviet collapse and thus not seen as comparable by most analysts). Rather,

individual single-party regimes, especially Mexico, have tended to be treated as unique

exceptions to the democratic rush of the third wave. Mexico specialists failed to notice that

single-party regimes in other parts of the world also survived for many decades despite economic

crises, and Asianists failed to notice that single-party regimes outside the areas of Chinese

settlement have resisted democratization as long as those in which Confucian values prevail.

The data analysis here puts the experiences of Mexico, Taiwan, Singapore, Malaysia, and

several Islamic countries in a useful comparative perspective. It shows that single-party regimes

are, on average, simply remarkably long-lived and resilient. They survive longer in richer

countries than in poorer, but even in Africa their longevity has been notable.

26

When facing intense pressure from donors or uncontrollable popular opposition, however,

single-party regimes usually negotiate their extrications. Like officers, the cadres of single-party

regimes can expect life as they know it to continue after liberalization or even democratization. If

they cannot avoid regime change, they are better off in a democracy than in some other form of

authoritarianism. Previously hegemonic parties have remained important in political life in most

countries that have fully democratized, but they have been outlawed and repressed in several

that did not. Consequently, leaders of dominant parties have good reason to negotiate an

extrication rather than risking a more violent ouster.

At the beginning of this paper I argued that the dearth of theories of authoritarian politics had

impeded the understanding of transitions. This study has taken a small step in the direction of

building such theories. The data collected for this study have made possible a few generalizations

about the likelihood of authoritarian breakdown that had remained hidden from analysts whose

studies focused on only a few cases or a single region.18 Much more remains to be done both of

theory building and data analysis.

18The data collection, however, would not have been possible without these prior studies. The kind of subtle and detailed evidence needed to test the hyptheses implied by the argument above could not have been found in the work of non-area specialists, and the hyptheses could never have been tested using off-the-shelf data sets.

27

References

Barros, Alexandre S.C. 1978. “The Brazilian Military: Professional Socialization, Political Performance and State Building.” PhD thesis. University of Chicago. 439 pp.

Bienen, Henry. 1978. Armies and Parties in Africa. New York: Africana Publishing. 278 pp.

Bollen, Kenneth. 1979. “Political Democracy and the Timing of Development.” American Sociological Review 44: 572-87

Bratton, Michael, and Nicolas van de Walle. 1997. Democratic Experiments in Africa: Regime Transitions in Comparative Perspective. Cambridge: Cambridge University Press. 307 pp.

Burkhart, R., and Michael Lewis-Beck. 1994. "Comparative Democracy: The Economic Development Thesis. American Political Science Review 88:903-10

Casper, Gretchen, and Michelle Taylor. 1996. Negotiating Democracy. Pittsburgh: Pittsburgh University Press. 287 pp.

Colomer, Josep. 1995. Game Theory and the Transition to Democracy: The Spanish Model. Aldershot, England: Edward Elgar, 134 pp.

Decalo, Samuel. 1976. Coups and Army Rule in Africa: Studies in Military Style. New Haven: Yale University Press. 284 pp.

Finer, Samuel. 1975. The Man on Horseback: The Role of the Military in Politics, 2nd ed. Harmondsworth, UK: Penguin. 305 pp.

Fontana, Andrés. 1987. “Political decision-making by a military corporation: Argentina, 1976-83.” PhD thesis. University of Texas. 218 pp.

Geddes, Barbara. 1999. "What Do We Know about Democratization after Twenty Years?" Annual Review of Political Science II, 115-44.

Geddes, Barbara. 2003. Paradigms and Sand Castles: Theory Building and Research Design in Comparative Politics, University of Michigan Press.

Haggard, Stephan and Robert Kaufman. 1995. The Political Economy of Democratic Transitions. Princeton: Princeton University Press. 391 pp.

Huntington, Samuel. 1991. The Third Wave: Democratization in the Late Twentieth Century. Norman: University of Oklahoma Press. 366 pp.

Jackman, Robert W. 1973. “On the Relations of Economic Development to Democratic Performance. American Journal of Political Science 17:611-21.

Janowitz, Morris. 1960. The Professional Soldier: A Social and Political Portrait. Glencoe: Free Press. 464 pp.

Janowitz, Morris. 1977. Military Institutions and Coercion in the Developing Nations. Chicago: University of Chicago Press. 211 pp.

Levi, Margaret. 1988. Of Rule and Revenue. Berkeley: University of California Press.

Linz, Juan and H. E. Chehabi, eds. 1998. Sultanistic Regimes. Baltimore: Johns Hopkins University Press. 284 pp.

Linz, Juan, and Alfred Stepan. 1996. Problems of Democratic Transition and Consolidation: Southern Europe, South America, and Post-Communist Europe. Baltimore: Johns Hopkins University Press. 479 pp.

Londregan, John, and Keith Poole. 1990. “Poverty, the Coup Trap, and the Seizure of Executive Power.” World Politics 42:151-83

Londregan, John, and Keith Poole. 1996. "Does High Income Promote Democracy? World Politics 49:1-30.

28

Nordlinger, Eric. 1977. Soldiers in Politics: Military Coups and Governments. Englewood Cliffs, NJ: Prentice-Hall. 224 pp.

North, Douglass. 1981. Structure and Change in Economic History. New York: W.W. Norton.

North, Douglass. 1989. "Institutions and Economic Growth: An Historical Introduction." World Development 17: 1319-32.

North, Douglass. 1990. Institutions, Institutional Change and Economic Performance. Cambridge: Cambridge University Press series on the Political Economy of Institutions

O'Donnell, Guillermo, and Philippe Schmitter. 1986. Transitions from Authoritarian Rule: Tentative Conclusions about Uncertain Democracies. Baltimore: Johns Hopkins University Press. 81pp.

Przeworski, Adam, Michael E. Alvarez, José Antonio Cheibub, and Fernando Limongi. 2000. Democracy and Development: Political Institutions and Well-Being in the World, 1950-1990. Cambridge: Cambridge University Press.

Przeworski, Adam, and Fernando Limongi. 1997. “Modernization: Theories and Facts.” World Politics 49:155-83.

Snyder, Richard. 1998. “Paths Out of Sultanistic Regimes: Combining Structuralist and Voluntarist Perspectives.” In Juan Linz and H. E. Chehabi, eds, Sultanistic Regimes. Baltimore: Johns Hopkins University Press. 284 pp.

Stepan, Alfred. 1971. The Military in Politics: Changing Patterns in Brazil. Princeton: Princeton University Press. 312 pp.

Valenzuela, Arturo. 1978. The Breakdown of Democratic Regimes: Chile. Baltimore: Johns Hopkins University Press. 140 pp.

Van de Walle, Nicolas, and Kimberly S. Butler. 1999. “Political Parties and Party Systems in Africa’s Illiberal Democracies,” unpublished manuscript, Michigan State University.

29

Figure 1

Games between Factions in Military and Single-Party Regimes Panel A: Game between Military Factions Majority Faction intervene barracks intervene a, b c, d

Minority Faction

barracks d, c b, a

Panel B: Game between Factions in Single-Party Regimes Majority

(Leader's Faction) in office out of office in office a, a b, c

Rival Faction

out of office c, b d, d

Where a>b>c>d, for both factions in both games

30

Figure 3: The Effect of Economic Performance on Stability in Different Kinds of Dictatorship Panel A

0.2

.4.6

.81

Pred

icte

d P

roba

bilit

y of

Fai

lure

: Pr(

Y=1)

-20 -10 0 10 20Growth Rate

lo medhi

Hi and Lo represent the 90% CI of the Predicted Probability

Economic Growth in Military Regimes

Middle line is the predicted probability of breakdown, given rates of growth. Top and bottom lines show the upper and lower bounds of a 90% confidence interval. Panel B: Effect of Economic Performance in Single-Party Regimes

0.2

.4.6

lo/m

ed/h

i

-20 -10 0 10 20Growth

lo medhi

Middle line is the predicted probability of breakdown, given levels of growth. Top and bottom lines mark the upper and lower bounds of 90% confidence interval.

31

Figure 2, Panel C:

0.2

.4.6

.8P

redi

cted

Pro

babi

lity

of F

ailu

re: P

r(Y

=1)

-20 -10 0 10 20Growth Rate

Military SPartyTriple

Median Predicted Probablity Show n

Economic Growth in Authoritarian Regimes

Top line is predicted probability of breakdown at given growth rate for military regimes. Middle line is predicted probability of breakdown for single-party regimes. Bottom line, triple hybrids.

For all graphs, control variables are set as follows: region is set at Central America (the “average” region); civil war, no; foreign imposed, no; all other variables are set at their means.

32

Table 1: Effect of Regime Type on Likelihood of Breakdown Weibull Regression,

Hazard Ratios

Multiple Imputation Logit, Coefficientsi

Regime Type:ii Military 3.62** 3.79** 1.20** 1.31** Military-Personalist 2.16** 1.64 0.57* 0.33 Single-Party Hybrid 1.05 1.22 0.11 0.28 Single-Party 0.33** 0.32** -1.19** -1.18** Triple Hybrid 0.17** 0.15** -1.82** -1.91** Level Developmentiii 0.67** 0.65** -0.49** -0.54** Growth Rateiv 0.94** 0.94** -0.07** -0.07** Foreign Imposed 1.77 2.33 0.81 1.00 Region:v Asia 1.77 1.34 0.16 -0.20 Central Am., Caribbean 1.41 1.34 0.15 0.09 Eastern Europe 0.82 0.67 -0.28 -0.45 Middle East 1.81 0.66 0.43 -0.61 North Africa 1.07 0.34 -0.33 -1.59** South America 4.02** 4.74** 1.10* 1.28** Sub-Saharan Africa 1.32 0.94 -0.17 -0.54 Percent Islam 1.01** 0.01** Civil War 1.57 0.46 Spline 1 0.01 0.02 Spline 2 0.06** 0.03** Spline 3 0.06 0.07* Spline 4 -0.02 0.14 Constant 0.63 0.79 Number of Observations

2118

2118

2118

2118

p 1.92 1.97 Percent Accurately Predicted

94.8

94.8

** Statistically significant at .05 or better. * Statistically significant at .1 to .05. 1 TSCS Logit, standard errors adjusted for clustering on coreg. 1 The left out regime category is Personalist. 1 Natural log of GDP per capita from Penn World Tables 2000, with missing values imputed using Amelia. Standard errors for this variable underestimated in Weibull models. 1 From Penn World Tables 2000, with missing values imputed, growth averaged over time T-1 and T-2. Standard errors for growth underestimated in the Weibull models. 1 Southern Europe is the left out region.

33

Table 2: Effect of Economic Performance on the Likelihood of Breakdown in Different Types of Authoritarian Regime

Logit

Coefficients P>|z|vi

Regime Type:vii Military 1.39 .000 Military-Personalist 0.34 .305 Single-Party Hybrid 0.28 .371 Single-Party -1.25 .001 Triple Hybrid -2.02 .000 Level Developmentviii -0.55 .001 Growth Rateix -0.03 .342 Foreign Imposed 1.16 .130 Region:x Asia 0.06 .926 Central Am., Caribbean 0.22 .699 Eastern Europe -0.38 .565 Middle East -0.34 .687 North Africa -1.36 .081 South America 1.49 .026 Sub-Saharan Africa -0.35 .571 Percent Islam 0.01 .001 Civil War 0.47 .093 Military-Growth Interaction -0.10 .107 Single-Party-Growth Interaction -0.09 .128 Triple Hybrid-Growth Interaction -0.02 .820 Spline 1 0.02 .380 Spline 2 0.07 .006 Spline 3 0.08 .282 Spline 4 -0.02 .870 Constant 0.64 .661 Number of Observations

2118

1 Standard errors adjusted for clustering on coreg. 1 The left out regime category is Personalist. 1 Natural log of GDP per capita from Penn World Tables 2000, with missing values imputed using Amelia. 1 From Penn World Tables 2000, with missing values imputed, growth averaged over time T-1 and T-2. 1 Southern Europe is the left out region.

i TSCS Logit, standard errors adjusted for clustering on coreg. ii The left out regime category is Personalist. iii Natural log of GDP per capita from Penn World Tables 2000, with missing values imputed using Amelia. Standard errors for this variable underestimated in Weibull models. iv From Penn World Tables 2000, with missing values imputed, growth averaged over time T-1 and T-2. Standard errors for growth underestimated in the Weibull models.

34

v Southern Europe is the left out region. vi Standard errors adjusted for clustering on coreg. vii The left out regime category is Personalist. viii Natural log of GDP per capita from Penn World Tables 2000, with missing values imputed using Amelia. ix From Penn World Tables 2000, with missing values imputed, growth averaged over time T-1 and T-2. x Southern Europe is the left out region.

35